In-house FIV tests- do people do a confirmatory test on +ves?

Re: In-house FIV tests- do people do a confirmatory test on +ves?

Yantha Smyth — Thu, 09 May 2013 22:08:47 GMT

Thanks for your input, and an update on the cat-

The practice did feel that confirmatory testing on the cat was not justified, and it was euthanased based on the in-house test kit result. I agree that the likelihood of this particular cat being a false positive would be tiny.

Re: In-house FIV tests- do people do a confirmatory test on +ves?

Martin Atkinson — Wed, 08 May 2013 15:20:22 GMT

I would have more concerns with a cat fitting the OP's description that was FIV negative. But the probability of a such a stereotype turning up a false positive is so low as to be negligible. If the cat is un-homeable what is the point in taking a chance or wasting any more money?

Re: In-house FIV tests- do people do a confirmatory test on +ves?

Bob Russell — Wed, 08 May 2013 13:26:08 GMT

[quote user="Hannah Wynne Richards"]

Bad luck if you're one of the cats unnecessarily killed

Wynne

[/quote]

We are talking mainly about rescue/feral cats that are likely be difficult to home as possible +ve's or a potential threat to the cat population if genuinely +ve. Sadly there is a big surplus of these already and limited resources.

Quite tough for a minority but common sense for the majority, I would have thought.

An in-house +ve pet or one that is a very good candidate for re-homing may be the exceptions to the rule.

Re: In-house FIV tests- do people do a confirmatory test on +ves?

Hannah Wynne Richards — Wed, 08 May 2013 11:33:48 GMT

Bad luck if you're one of the cats unnecessarily killed

Wynne

Re: In-house FIV tests- do people do a confirmatory test on +ves?

Rob Loxley — Mon, 06 May 2013 20:32:34 GMT

I find the ABCD guidelines a good reference

Re: In-house FIV tests- do people do a confirmatory test on +ves?

John Flynn — Mon, 06 May 2013 19:21:03 GMT

[quote user="David Mills"]Your calculations are sound, but the extrapolation to killing healthy cats is not. Yes,10%, on your calculations is correct. The problem, I think, is with the limitations of applying sensitivity and specificity to populations both in terms of the prevalence of disease in the population, and the population that is presented to you because sens/spec are calculated from different populations to your own, and often from small numbers. There are also factors not considered in trials which are very important. That is why I don't think the 10% is representative - yes its a deductive logical conclusion - but its a simplification that almost entirely ignores population factor. [/quote]

I don't know I totally follow where I've extrapolated un-soundly with respect to killing non-FIV cats (but appreciate I may have and can tell from your posts that I know far less than you do about the epidemiology of FIV both practically and theoretically).

The only application of sensitivity would be to suggest that the true-prevalence in your area will be higher than your test-prevalence, by how much will depend on the sensitivity of the test in the population being tested. If the test was very insensitive (like the bovine TB comparative intradermal skin test I'm more familiar with), then this could be relevant, as your disease prevalence would be, say, 25% instead of your test-positive prevalence of 20%. Even at such a poor sensitivity, this would have minimal effect on the rough 10% figure calculated already (if the sensitivity was extremely low, then the disease-prevalence would be much higher than the 20% test-positive prevalence you see and therefore the chances that a test-positive cat that was euthanased had FIV would go up, but only because the chances of any given cat being test-positive were much higher pre-test and a test with that poor a sensitivity wouldn't be worth wasting the time doing it in the circumstances you describe).

The specificity in your clinic situation is only likely to be lower than the 98% (which I assume you get from the manufacturers); the specificity is never going to be higher than this in your test-population, so I can't think of a factor relating to the specificity of the test that would lower the 10% estimate.

[quote user="David Mills"]We currently do test everything, but whether this is useful or not we don't know because we cannot afford to PCR-verify.[/quote]

Another determinant of whether this is useful (as well as what resources are saved by euthanasing FIV-positive cats that may require more treatment and be difficult to re-home) is whether such a policy has a measureable effect on the local disease prevalence. I've never done test-and-cull for FIV, but I know how difficult BVD, Johnes or other test-and-cull programmes can be even in quite a controlled population like a pedigree farm tested yearly. I do wonder how much of an effect on FIV prevalence locally this policy has, but suspect much less than your presumed policy of neutering strays (which will decrease the population density).

[quote user="David Mills"]I have no idea, but I struggle to see how any antibody test can be 100% specific.[/quote]

It can't. But the difference between 98% and 99.9% specificity in your setting would be the difference between 1 in 10 FIV-test-positive cats and 1 in 200 FIV-test-positive cats being euthanased without actually having FIV antibodies, or between 20 cats without FIV and 1 cat without FIV being euthanased a year. I'm not saying that this matters enough for the extra inconvenience and use of resources etc, but it is a definable difference.

I'm not disagreeing with your policy per se, but I do think that phrasing a figure as in number of test positives that don't have the disease helps to make a 2% false positive rate come to life a bit, in this case (with a test-positive prevalence of about 20%, a figure that you can either reliably estimate or check in detail with a simple tally chart to record results in the test area) that comes to about 10% of test-positives.

Re: In-house FIV tests- do people do a confirmatory test on +ves?

David Mills — Mon, 06 May 2013 18:49:26 GMT

[quote user="John Flynn"]

I understand that making general assumptions re prevalence can be very misleading, but I thought you had a pretty good idea what the test-positive prevalence was in the population you are seeing (1 in 4 or 5 cats you said), and if you're doing 3 or 4 tests a day then I don't see what is wrong with following your numbers through like this:

That is approximately 1000 cats tested for FIV daily.

You have said that roughly 1 in 5 of these are positive.

You have said that you expect a 2% false positive rate (and I've taken that to mean that you expect a 98% specificity which doesn't sound unreasonable)

Of the 1000 cats tested per year then:

200 will test positive

20 false positives will occur

that is about 1 in 10 cats, or 20 cats a year without FIV euthanased under this policy - I don't think that's a bad thing necessarily, but I don't see why you don't believe the 10% figure - it's based on the estimates you made in your clinical setting? You could easily measure the number of positive results you get and record whether the cat was sick or healthy if you want to see how that figure varies in your own setting - obviously it's going to depend on how many sick/healthy cats you test and I don't know your policy on this [/quote]

Your calculations are sound, but the extrapolation to killing healthy cats is not. Yes,10%, on your calculations is correct. The problem, I think, is with the limitations of applying sensitivity and specificity to populations both in terms of the prevalence of disease in the population, and the population that is presented to you because sens/spec are calculated from different populations to your own, and often from small numbers. There are also factors not considered in trials which are very important. That is why I don't think the 10% is representative - yes its a deductive logical conclusion - but its a simplification that almost entirely ignores population factor.

We currently do test everything, but whether this is useful or not we don't know because we cannot afford to PCR-verify.

[quote user="John Flynn"]A test for FIV antibodies from Glasgow (we use Langford, I don't have their price list to hand) will cost £18 and have a significantly higher specificity than the SNAP test I would have thought.[/quote]

I have no idea, but I struggle to see how any antibody test can be 100% specific.

[quote user="John Flynn"]With respect to your own situation, are the numbers of FIV positives coming down or do you think that the test and cull policy is just maintaining the prevalence at the current levels but would otherwise increase?[/quote]

The latter. Probably.

Re: In-house FIV tests- do people do a confirmatory test on +ves?

John Flynn — Mon, 06 May 2013 18:20:05 GMT

That is approximately 1000 cats tested for FIV yearly.

You have said that roughly 1 in 5 of these are positive.

You have said that you expect a 2% false positive rate (and I've taken that to mean that you expect a 98% specificity which doesn't sound unreasonable)

Of the 1000 cats tested per year then:

200 will test positive

20 false positives will occur

In the setting in which I work it is obviously different.

SNAP tests cost us way more than £10 a pack (I don't know how many are in a pack), but then we would be purchasing singly and nowhere near 1000 a year. Are you sure they are SNAP tests (I think these are manufactured by Idexx)?

A test for FIV antibodies from Glasgow (we use Langford, I don't have their price list to hand) will cost £18 and have a significantly higher specificity than the SNAP test I would have thought. The cost of sending for FIV/FeLV to an external laboratory works out surprisingly close cost-wise (last time I checked) in our setting, especially when the need to stock fridge space with tests that aren't used at anything close to your frequency is considered.

When it comes to strays, we only FIV test sick ones (we don't really take in healthy strays). Normally we hold on to strays for at least a few days to see if an owner will come forward, so wouldn't be euthanasing immediately anyway. Feral cats brought in for neutering are neutered and released again.

An FIV positive stray, without an owner coming forward, is either euthanased or re-homed in a single-cat indoor situation (if that would be reasonable for the cat in question and a home can be found within a week) or re-homed to a home with other FIV-positive cats in an indoor situation.

I have no ambition to fund an FIV test-and-cull programme in the area personally, and am unaware of anyone else funding such widescale testing.

With respect to your own situation, are the numbers of FIV positives coming down or do you think that the test and cull policy is just maintaining the prevalence at the current levels but would otherwise increase?

Re: In-house FIV tests- do people do a confirmatory test on +ves?

David Mills — Mon, 06 May 2013 17:05:50 GMT

[quote user="John Flynn"]

[quote user="David Mills"]As a charity, as positives we see are euthanased, largely because we see a lot of strays and approximately 1 in 4-5 are positive - this is well above the false positive level that can be expected, and although we probably euthanase the occasional non-carrier, the risk of missing these, and the financial burden of retesting positives would be disastrous and ruinaous respectively.[/quote]

Fair enough, but on the 2% false positive rate you quote (which assumes test performed correctly etc etc), I'm assuming this means 2% of non-FIV cats test positive, then:

50 cats tested:

10 have FIV (based on your suggested prevalence of 1 in 4-5cats)

1 without FIV tests positive

therefore 1 out of every 11 cats that you euthanase based on a positive FIV test didn't have FIV.

I don't know how many cats you euthanase a year, but I'd say that adds up to a fair few cats, approximately 10% of those testing positive for FIV in your setting unless I'm mistaken?

That's fine if that's the policy, and I understand why in such a setting getting an instant result has big benefits, but I tend to not use SNAP tests (quite expensive to start with) and just send to an external laboratory where the result will be more valid first time round.

[/quote]

You can't just extrapolate the numbers like that because there are so many variables; to do so reduces a complex argument to percentages. Perhaps I didn't make that clear.

- on current estimates prevalence in sick cats is 12-18%. Prevalence in non-sick cats is 3-6%. These are mainly from cats presented to a vets - we have no (real) idea the prevalence in the stray population. Even so, a sick cat is far more likely to have FIV than a non-sick cat. Therefore the PPV (which uses both sen/spec and prevalence) for the SNAP test is greatly increased for a sick, stray (and, as it goes, male (neutering no significant effect)) cat.

- some cats likely carry the antibody without the virus, or very closely related antibodies that cross react with the test, or antibodies that remain from MDA - no one knows when they finally (if ever) leave the system. Recently, transmission of the virus has shown to be incomplete. Some cats may be transiently infected. The only way, truly, of seeing which are carriers is to PCR test TWICE - once at admission and then 3-4(?who knows) weeks later/

- it depends where you get your SNAP test from. Ours are less than £10 per pack. We use about 3/4 a day. PCR is around £35 a pop.

- on your scheme of higher sensitivity, then, you're looking at £35 for a single test for one disease being used as a screening tool, with time lag in between results; double that if you want true confirmation. Meanwhile, the cat could be out roaming again, spreading disease (or not) or spreading within household etc. That to me is irresponsible in terms of the cat's and population's welfare, especially when there is a cat-side test which can give you a pretty good idea - in a sick cat - of whether FIV is there. And what about the well cats? As many cats can live normal lives with FIV. Or do you PCR everything? Do you PCR all sick cats? Or just some? How do you decide? Who pays? You would keep a stray cat for the necessary 3-4d to wait for a PCR result to come back? Again, who pays?

- Yes, we likely do euthanase healthy cats without FIV. I don't believe it is anywhere near 10%. Even if it was, I fully support this. The disease cannot - indeed should not - be considered as a problem just in the individual (like any infectious disease, arguably). There is a risk analysis to be made on behalf of the rest of the population - whether this be household, hospital, rehoming centre, or local stray population. Any reasonable ethical analysis of the situation would conclude that euthanasia in these circumstances is the correct option - indeed, even if the healthy rate was higher, I think it would still be justified.

Re: In-house FIV tests- do people do a confirmatory test on +ves?

John Flynn — Mon, 06 May 2013 16:21:41 GMT

Fair enough, but on the 2% false positive rate you quote (which assumes test performed correctly etc etc), I'm assuming this means 2% of non-FIV cats test positive, then:

50 cats tested:

10 have FIV (based on your suggested prevalence of 1 in 4-5cats)

1 without FIV tests positive

therefore 1 out of every 11 cats that you euthanase based on a positive FIV test didn't have FIV.

I don't know how many cats you euthanase a year, but I'd say that adds up to a fair few cats, approximately 10% of those testing positive for FIV in your setting unless I'm mistaken?

PS - I would agree to euthanase the cat in question above though.

Re: In-house FIV tests- do people do a confirmatory test on +ves?

David Mills — Mon, 06 May 2013 15:18:29 GMT

I don't have the figures to hand but with the SNAP you will, for FIV or FeLV, get a false positive roughly 2% of the time.

However, IF the cat is ill for any reason, incl CBAs etc, then the accuracy is very good.

As a screening test it can be useful, but is heavily dependent on your area (ergo the prevalence of FIV/FeLV - wheter anyone knows the true prevalence in an area is itself contentious - many of the pronunciations on this are nothing but anecdotal, referral centre opinion); if there are a lot of strays and your clinic is seeing a lot of strays (especially ill ones/ill enough to be caught) then I think it is adequate. However if you don't see a lot of strays in the area, and see healthy 'garden cats' or similar, then the number of false positives will greatly outweigh the true positives, and it is therefore not a good test. Add into this mix the fact many cats (healthy and ill) can be transiently infected, and then self-cure, and it all becomes quite messy.

As a charity, as positives we see are euthanased, largely because we see a lot of strays and approximately 1 in 4-5 are positive - this is well above the false positive level that can be expected, and although we probably euthanase the occasional non-carrier, the risk of missing these, and the financial burden of retesting positives would be disastrous and ruinaous respectively.

In your situation, I would euthanase

Re: In-house FIV tests- do people do a confirmatory test on +ves?

Virginia Campbell — Mon, 06 May 2013 13:19:45 GMT

We do a lot of Cats Protection work. Their protocol in this case would be to euth the cat.

(wildie and snap positive sickie. If he was tame enough to be potentially rehomable, they would send confirmation tests).

he's a feral entire male with fight wounds and is ill...I believe the snap.

If he's that wild, I wouldn't think it fair to bung him in a cage till you can get confirmation blood results back....you know what they are going to be.

Re: In-house FIV tests- do people do a confirmatory test on +ves?

James Laidlaw — Mon, 06 May 2013 11:34:51 GMT

I can't explain all the mathematics involved, but due to fiv having a low prevalence, there is a higher chance of a false positive with the in house snap test so if euthanasia is to be considered then you really should confirm with PCR/external lab. However, in this situation with a stray that might change some people's opinions on how to proceed.